List your company

Visual Probing on VisualProbe Medium

52.9AccuracyMini-o3
10.36421.40732.4543.493Jun 11, 2026

Evaluation results

MethodLinks
Mini-o3Model Type=Tool-Augmented2026.0652.9
PERIA-8BModel Type=Proposed2026.0641.2
Gemini-2.5-ProModel Type=Commercial2026.0638.7
w/o RLModel Type=Ablation2026.0637.3
Gemini-2.5-FlashModel Type=Commercial2026.0636
Qwen3-32BModel Type=Open-Source2026.0632
Qwen3-235BModel Type=Open-Source2026.0632
Qwen3-8BModel Type=Backbone Baseline2026.0631.3
Qwen3-30BModel Type=Open-Source2026.0630.2
VTool-R1-7BModel Type=Tool-Augmented2026.0626.4
InternVL3.5-8BModel Type=Open-Source2026.0625.7
R1-OneVision-7BModel Type=Tool-Augmented2026.0621.7
GPT-5Model Type=Commercial2026.0621.6
w/o ToolModel Type=Ablation2026.0621.2
w/o SFTModel Type=Ablation2026.0618.6
InternVL3.5-14BModel Type=Open-Source2026.0612